Table 3.
ROC curve analyses for the diagnosis of DLB.
| Predictor | AUC (95% CI) | Threshold | Specificity (%) | Sensitivity (%) | P value |
|---|---|---|---|---|---|
| DAT-AP | 0.75 (0.66–0.85) | 0.60 | 87.8 | 56.4 | 0.092 |
| DAT-PP | 0.79 (0.79–0.88) | 0.52 | 79.6 | 72.7 | Reference |
| FDR-PP | 0.82 (0.74–0.91) | 0.59 | 89.8 | 69.1 | 0.225 |
| DAT-AC | 0.83 (0.75–0.91) | 0.65 | 91.8 | 69.1 | 0.241 |
| DAT-PC | 0.85 (0.77–0.92) | 0.71 | 93.9 | 65.5 | 0.119 |
| DAT-PC + FDR-PP | 0.88 (0.81–0.94) | 0.54 | 85.7 | 81.8 | < 0.001 |
| PC1 | 0.98 (0.96–0.999) | 0.83 | 100 | 83.6 | < 0.001 |
| DAT-PP + PC1 | 0.98 (0.96–0.999) | 0.81 | 98.0 | 85.5 | < 0.001 |
| DAT-AP + PC1 | 0.98 (0.96–0.999) | 0.83 | 100 | 83.6 | < 0.001 |
| DAT-PC + PC1 | 0.98 (0.96–0.999) | 0.22 | 87.8 | 96.4 | < 0.001 |
| DAT-AC + PC1 | 0.98 (0.96–0.999) | 0.78 | 95.9 | 87.3 | < 0.001 |
| FDR-PP + PC1 | 0.99 (0.97–1.00) | 0.30 | 91.8 | 96.4 | < 0.001 |
Results correspond to ROC curve analyses for the diagnosis of DLB. P values are the results of analyses based on DeLong’s method comparing each model’s accuracy with that of the model using DAT-PP as a predictor.
AUC area under the curve, DAT dopamine transporter, DAT-AC DAT uptake in the anterior caudate, DAT-AP DAT uptake in the anterior putamen, DAT-PC DAT uptake in the posterior caudate, DAT-PP DAT uptake in the posterior putamen, DLB dementia with Lewy bodies, FDG 18F-fluorodeoxyglucose uptake, FDR-PP FDG to DAT ratio in the posterior putamen, PC1 the first principal component of FDG subject residual profile, ROC receiver operating characteristic.